Consanguinity: A Risk Factor for Preterm Birth at Less Than 33 Weeks’ Gestation

Abstract

Consanguinity promotes homozygosity of recessive susceptibility gene variants and can be used to investigate a recessive component in diseases whose inheritance is uncertain. The objective of this study was to assess the association between consanguinity and preterm birth (PTB), stratified by gestational age and clinical presentation (spontaneous vs. medically indicated). Data were collected on 39,745 singleton livebirths without major birth defects, admitted to 19 hospitals in Lebanon, from September 2003 to December 2007. Deliveries before completed 33 weeks’ gestation and deliveries at 33–36 weeks’ gestation were compared, with respect to cousin marriage, with those after completed 36 weeks’ gestation by using multinomial multiple logistic regression. Overall, infants of consanguineous parents had a statistically significant 1.6-fold net increased risk of being born at less than 33 weeks’ gestation compared with infants of unrelated parents. This association was statistically significant only with spontaneous PTB. There was no increased risk of being born at 33–36 weeks’ gestation associated with consanguinity for both clinical presentations of PTB. Our findings support a genetic contribution to early onset PTB and suggest that early PTB should be targeted in future genetic studies rather than the classic lumping of all births less than 37 weeks’ gestation.

With approximately 13 million annual births before 37 weeks’ gestation worldwide, preterm birth (PTB) is a major obstetric and neonatal challenge (1). It remains the leading cause of neonatal mortality globally, accounting for 27% of all direct causes of deaths (2). Millennium Development Goal 4 aims at reducing the under-5 child mortality by two thirds by 2015. With 38% of these deaths occurring in the neonatal period, achieving this goal will not be feasible unless significant action is taken for the prevention of PTB globally (2). Despite extensive research, causes of PTB remain an unresolved issue in obstetric care (3). Recent evidence points to a multifactorial etiology of PTB, with interplay of environmental and genetic factors. Although some environmental risk factors have long been recognized, the genetics of PTB is lately receiving worldwide attention from the scientific community (4, 5).

Consanguinity, the marriage between relatives, is often used to investigate a recessive component in diseases, because it favors the reemergence of deleterious alleles that run in families (6). Exploring the association of consanguineous marriages with PTB can give important insights as to the role of a genetic component in PTB. Previous studies have extensively investigated an association between consanguinity and lower birth weight (7–17), but few have attempted to explore its association with PTB (10, 12–14, 18–20). The majority of these studies have methodological limitations, including selection bias (12, 14), failure to control for potential confounding factors (10, 12–14), and small sample size (10, 12, 13, 19).

In Lebanon, marriage of relatives is a common cultural practice and provides the opportunity to study the effect of consanguinity on PTB. In this paper, and using a large data set from a perinatal neonatal network in Lebanon, we investigate the association between consanguinity and PTB, stratified by different levels of gestational age and clinical presentation, whether spontaneous or medically indicated.

MATERIALS AND METHODS

Study design

This study was a cross-sectional analysis on newborns admitted as part of the National Collaborative Perinatal Neonatal Network to 19 hospitals in Lebanon from September 1, 2003, to December 31, 2007. Trained personnel collect data on a daily basis pertaining to all deliveries at participating centers. Data sources include abstraction from medical charts and direct interviews with mothers. The study was approved by the Institutional Review Board at the American University of Beirut.

Study variables

Data on parental consanguinity were collected through direct interviews with mothers in the immediate postpartum period before hospital discharge. Parental consanguinity was categorized into unrelated, first cousins, and second cousins. Gestational age was established on the basis of an early first trimester ultrasound in 91% of the cases and the last menstrual period in the remaining cases. It was categorized into 3 levels: deliveries before completed 33 weeks’ gestation (early PTBs), deliveries at 33–36 weeks’ gestation (late PTBs), and deliveries after completed 36 weeks’ gestation (reference category: term births). PTBs were classified as spontaneous if they were preceded by spontaneous labor or rupture of membranes.

The sociodemographic characteristics considered as potential confounders were maternal and paternal age, maternal education, working status during pregnancy, and household crowding index. Maternal characteristics included parity, prepregnancy body mass index, maternal height, weekly weight gain rate, and cigarette and narghile smoking. Information on pregnancy complications was available and included bleeding, hypertensive disorders, gestational diabetes, amniotic fluid abnormalities, and utilization of assisted reproductive techniques. The gender of the newborns was also considered as a potential neonatal confounder.

Study population

During the study period, a total of 54,248 liveborn infants were delivered at participating hospitals. Newborns were excluded if they were multiple gestations (n = 2,409), had major birth defects (n = 1,195), or had missing data on consanguinity (n = 3,422), gestational age (n = 1,373), type of gestation (single or multiple) (n = 602), birth defects (n = 3,733), or clinical presentation of the delivery (n = 3,488). The final sample consisted of a total of 39,745 singleton livebirths. Among this sample, there were only 819 women (2.1%) who delivered more than once during the study period giving birth to 1,648 babies. Exclusion of these cases or adjustment for them in the analysis did not impact the final results and, thus, they were included in the final sample.

Analysis

Analyses were done by using SPSS, version 15 (SPSS, Inc., Chicago, Illinois), and STATA, version 10.0 (StataCorp LP, College Station, Texas), software. Cross-tabulation between consanguinity and the different covariates was performed in order to define the characteristics of the studied population. The association between PTB and consanguinity and the other covariates was assessed by using the χ² test. Crude odds ratios and 95% confidence intervals were calculated by using bivariate multinomial regression analysis with the baseline category being term births. The independent effect of consanguinity on the risk of early and late PTB was assessed by performing multinomial multiple logistic regression analysis adjusting for within-hospital clustering of the data by using a robust estimator for the standard errors that allows correlation between responses within hospitals. The regression analysis included all variables that proved to be significantly associated with the outcome at the bivariate level (2-tailed P < 0.2). There was a concern that pregnancy complications might be intermediate outcomes of pregnancy; therefore, these covariates were not included in the final regression models. However, for further reassurance, regression analysis including these variables was carried out, and there was no noticeable change in the regression coefficients (not shown).

Because it is often argued that spontaneous and medically indicated PTBs have different risk factors (21, 22), the analyses were stratified by clinical presentation of PTB. In both cases, early and late PTBs were compared with the total term births (spontaneous and indicated).

RESULTS

Table 1 describes the characteristics of the study population by degree of consanguinity. Women in consanguineous unions were mainly Muslims, of lower socioeconomic status, had higher prepregnancy body mass index and lower pregnancy weight gain, and also smoked more than mothers in unrelated unions.

Table 1.

Characteristics of the Study Population by Consanguinity Level in Lebanon, Liveborn Singletons, Lebanon, 2003–2007

	Parental Consanguinity
Characteristics	Unrelated		First Cousins		Second Cousins		P Valuea
	No.	%	No.	%	No.	%
Religion
Muslim	24,273	75.8	3,431	96.8	1,251	96.2
Christian	7,756	24.2	115	3.2	50	3.8	<0.001
Maternal age, years
<20	1,242	3.6	208	5.5	88	6.3
20–34	26,944	78.4	2,954	77.9	1,095	78.7
≥35	6,173	18.0	631	16.6	208	15.0	<0.001
Paternal age, years
<25	924	2.8	182	4.9	62	4.6
25–44	30,527	91.4	3,321	89.5	1,222	89.8
≥45	1,931	5.8	209	5.6	77	5.7	<0.001
Maternal education
Illiterate to elementary	4,763	14.2	940	25.0	310	22.4
Intermediate/secondary	15,034	44.7	2,056	54.7	722	52.3
Technical/university	13,808	41.1	760	20.2	349	25.3	<0.001
Maternal working status
Housewife	25,257	74.8	3,391	90.5	1,221	88.6
Working	8,505	25.2	354	9.5	157	11.4	<0.001
Crowding index
<1	19,598	59.8	1,498	40.9	625	46.5
1–2	12,715	38.8	2,035	55.6	674	50.2
≥3	478	1.5	128	3.5	44	3.3	<0.001
Parity, no.
0	12,339	36.0	1,037	27.4	399	28.6
1–2	16,695	48.7	1,668	44.0	658	47.2
≥3	5,241	15.3	1,084	28.6	337	24.2	<0.001
Prepregnancy body mass index
Underweight	1,516	5.0	106	3.4	43	3.6
Normal	19,999	66.0	1,895	61.0	727	61.1
Overweight	6,910	22.8	870	28.0	324	27.2
Obese	1,885	6.2	235	7.6	95	8.0	<0.001
Maternal height, cm
<155	2,105	6.8	264	8.3	103	8.3
155–159	5,537	17.9	604	18.9	223	18.1
≥160	23,211	75.2	2,322	72.8	908	73.6	0.003
Weight gain rate, kg/week
<0.27	11,101	33.7	1,441	40.9	498	38.4
0.27–0.51	19,110	58.1	1,836	52.1	702	54.2
≥0.52	2,693	8.2	246	7.0	96	7.4	<0.001
Cigarette smoking
No	29,958	87.9	3,343	89.0	1,205	87.1
Before pregnancy	1,308	3.8	90	2.4	35	2.5
During pregnancy	2,809	8.2	322	8.6	143	10.3	<0.001
Narghile smoking
No	30,212	88.7	3,370	89.7	1,225	88.6
Before pregnancy	1,984	5.8	167	4.4	74	5.4
During pregnancy	1,879	5.5	218	5.8	84	6.1	0.010
Assisted reproductive techniques	329	1.0	35	1.0	8	0.6	0.375
Bleeding	2,032	5.9	220	5.9	85	6.2	0.918
Hypertensive disorders	503	1.5	58	1.5	24	1.7	0.680
Gestational diabetes	418	1.2	27	0.7	13	0.9	0.018
Amniotic fluid abnormalities	1,360	3.9	275	7.2	99	7.1	<0.001
Female sex	16,583	48.7	1,797	48.3	673	49.3	0.838

^aTwo tailed P value of the chi-square test.

Bivariate analysis showed that the proportion of first and second cousins was 15.4% and 4.6%, respectively, among early PTB compared with 9.6% and 3.5%, respectively, among term births, with the relationship being statistically significant only with first-cousin marriage (odds ratio = 1.7, 95% confidence interval (CI): 1.3, 2.4) (Table 2). Similar trends were observed when the analysis was stratified by spontaneous versus medically indicated PTB. No association was observed between consanguinity and any type of late PTB (Table 2).

Table 2.

Bivariate Association Between Consanguinity and PTB Stratified by Type of PTB, Liveborn Singletons, Lebanon, 2003–2007

Parental Consanguinity	Early PTB		Late PTB		Terma		Early PTB/Terma		Late PTB/Terma
	No.	%b	No.	%b	No.	%b	OR	95% CI	OR	95% CI
	Total PTB
Not related	224	80.0	1,544	87.9	32,774	86.9	1.0	Referent	1.0	Referent
First cousins	43	15.4	155	8.8	3,608	9.6	1.7	1.3, 2.4	0.9	0.8, 1.1
Second cousins	13	4.6	58	3.3	1,326	3.5	1.4	0.8, 2.5	0.9	0.7, 1.2
	Spontaneous PTB
Not related	180	80.0	1,247	88.1	32,774	86.9	1.0	Referent	1.0	Referent
First cousins	34	15.1	123	8.7	3,608	9.6	1.7	1.2, 2.5	0.9	0.7, 1.1
Second cousins	11	4.9	46	3.2	1,326	3.5	1.5	0.8, 2.8	0.9	0.7, 1.2
	Indicated PTB
Not related	44	80.0	297	87.1	32,774	86.9	1.0	Referent	1.0	Referent
First cousins	9	16.4	32	9.4	3,608	9.6	1.9	0.9, 3.8	1.0	0.7, 1.4
Second cousins	2	3.6	12	3.5	1,326	3.5	1.1	0.3, 4.6	1.0	0.6, 1.8

Abbreviations: CI, confidence interval; OR, odds ratio; PTB, preterm birth.

^aAll births >36 weeks, spontaneous and indicated.

^bColumn percentage.

In view of the small sample size in the subcategories of gestational age and consanguinity and because comparable measures of association for first and second cousins with PTB were observed in the bivariate analysis, first and second cousins were combined into one category in the multiple logistic regression. After controlling for the other risk factors and adjusting for within-hospital clustering, we found that cousin marriage remained significantly associated with spontaneous early PTB, with infants of these marriages having a 1.6-fold increased risk of being born spontaneously at less than 33 weeks’ gestation compared with infants of unrelated parents (95% CI: 1.1, 2.4). Consanguinity was associated with a similar increase in risk of medically indicated early PTB, but this did not reach statistical significance (odds ratio = 1.6, 95% CI: 0.7, 3.9). Even after adjustment for potential confounders, there was no association between consanguinity and late PTB, whether spontaneous or medically indicated (Table 3).

Table 3.

Multinomial Logistic Regression Modeling for the Independent Effect of Consanguinity on Early and Late PTB,^a Liveborn Singletons, Lebanon, 2003–2007

Parental Consanguinity	Early PTB		Late PTB
	OR	95% CI	OR	95% CI
	Total PTBb
Not related	1.0	Referent	1.0	Referent
First or second cousins	1.6	1.1, 2.4	1.0	0.8, 1.2
	Spontaneous PTBb
Not related	1.0	Referent	1.0	Referent
First or second cousins	1.6	1.1, 2.4	1.0	0.8, 1.2
	Indicated PTBc
Not related	1.0	Referent	1.0	Referent
First or second cousins	1.6	0.7, 3.9	1.0	0.7, 1.6

Abbreviations: CI, confidence interval; OR, odds ratio; PTB, preterm birth.

^aBaseline category: all births >36 weeks.

^bAdjusted for maternal age, paternal age, maternal education, working status, crowding index, prepregnancy body mass index, weekly weight gain, cigarette and narghile smoking before and during pregnancy, parity, assisted reproductive techniques, and newborn's sex. Standard errors were adjusted for clustering on hospital of delivery.

^cAdjusted for the same variables as in “all PTB” excluding working status during pregnancy, narghile smoking, and newborn's sex. Standard errors were adjusted for clustering on hospital of delivery.

As a check on whether consanguinity was an important risk factor for PTB, the final regression model performed for the total PTB was compared with a similar model containing all the covariates without consanguinity. Based on the likelihood ratio test, the contribution of consanguinity in predicting early PTB was statistically significant.

DISCUSSION

This work is the first investigation on the independent effect of consanguinity on the risk of PTB stratified by different levels of prematurity and clinical presentation. It showed that consanguinity is a risk factor for early but not late PTB, with the association being statistically significant only with spontaneous early PTBs. Overall, infants of consanguineous parents had 1.6-fold increased risk of being born at less than 33 weeks’ gestation compared with infants of unrelated parents (95% CI: 1.1, 2.4).

Previous studies on the effect of parental consanguinity on PTB have defined it at less than 37 weeks’ gestation and appear to be inconclusive. Two of the most recent investigations have reported a significant increase in the risk of PTB (18, 19). These are the few studies that controlled for potential confounders by using regression analysis. In the small case-control study of Saudi women, the reported odds ratio for the association between consanguinity and PTB was 3.0 (19), while the odds ratio of 1.8 reported in Jordan is more comparable to the findings of our study (18). A cross-sectional study conducted at one hospital in Lebanon also reported comparable odds ratios of 1.6 and 1.3 for PTB with first and second cousin marriage, respectively (12). Although this study and the remaining body of literature have failed to detect a significant association between consanguinity and PTB (10, 12–14), it remains unclear whether this was due to several methodological weaknesses, such as lack of control of covariates, or, in light of our results, whether this was due to the lumping of the early and late PTB in one outcome category.

A familial component to PTB has been recognized in the literature. Twin and family studies suggest that genetic effects explain, at least partially, the observed familiality (20, 23, 24). This pattern is demonstrated through strong patterns of recurrence of PTB. Studies on the heritability of gestational age among spontaneous deliveries at term have suggested that both maternal and fetal genes contribute to the normal variability in PTB (25, 26). However, recent studies in northern European populations restricted to PTB have shown that most heritable effects in PTB exert their effects through the mother, while a substantial role of the fetal genotype—underlying a paternal effect—is less obvious (27–29). Our results suggest that, although maternal genes might have a more important effect in determining PTB, genes inherited from the father should not be disregarded and that a classic polygenic mode of inheritance should not be overlooked in future genetic studies. The effect of paternal genes on PTB should be investigated with further scrutiny. As one possibility, there may be differences between populations in the role of maternal versus fetal genes.

One of the advantages of this study was the stratification by clinical presentation of PTB. Although much evidence has been building to split these groups in genetic epidemiologic studies (1), many argue that these 2 types share a common mechanism and should be aggregated (1). In our study, the odds ratio for the association between consanguinity and early PTB was similar for spontaneous and medically indicated PTB, but the association was statistically significant only with the former. Consanguinity was not a risk factor for any type of late PTB. It is worth noting that the sample of medically indicated PTB in our study was small, and these findings ought to be reproduced in larger investigations.

Despite the availability of information on many sociodemographic, behavioral, and medical confounders in the National Collaborative Perinatal Neonatal Network database, the possibility of bias due to residual confounding, whether from measurement errors in available covariates or from unmeasured confounding, cannot be fully eliminated. For example, our use of weekly weight gain by dividing the total pregnancy weight gain by the gestational age is not optimal because weight gain does not increase linearly with gestational age, with much of the gain occurring in the third trimester. A better measurement of this confounder could not be obtained with the available data. Information was also not available on intrauterine infections and bacterial vaginosis. However, the effect of residual confounding on exposure–outcome associations is complex, and it is not clear in which direction the potential bias would point (30). It is also unlikely that strong confounders might have been missed because these would usually be identified in the literature (31). Another limitation was that some cases were excluded for missing data on the main exposure, consanguinity. However, it is likely that this effect was only marginal, as there is no evidence that the missing information followed any specific pattern related to the exposure or outcome.

This multicenter study analyzed data on all deliveries at 19 major hospitals in Lebanon. The rate of PTB (<37 weeks’ gestation) in the original total sample (n = 54,248) was 8.2%, which is comparable to the reported global rate of 9.6% (32) and to the rate of 9.0% in Lebanon available from a national survey (33). The total sample represents around 17% of the total births in Lebanon for the specified period of time (http://www.cas.gov.lb). With over 94% of deliveries in Lebanon occurring in hospitals (34), we believe that our findings have a relatively good degree of generalizability. They indicate that consanguinity increases the risk of early and not late PTB. With consanguinity suggesting the involvement of a recessive mode of inheritance, our findings present a new insight to the mechanisms leading to PTB and suggest that future genetic studies should target specifically early PTB when investigating genetic origins of PTB.

Glossary

Abbreviations

CI

confidence interval

PTB

preterm birth